I recently started using Word Flow (predictive text), and it felt like I was actually typing one-handed messages faster than usual. However, was I really typing faster, or was it just a placebo effect? I put it to the test to find out.
|Normal 1-Hand||Predictive 1-Hand|
|10.63 seconds||18.47 seconds|
|33.43 seconds||56.17 seconds|
|26 seconds||41 seconds|
|13 seconds||25 seconds|
|23 seconds||25 seconds|
|11 seconds||24 seconds|
It quickly became clear that Word Flow’s text predictions were actually slowing me down. Typing normal with one hand was nearly twice as fast as typing one-handed using the predictions. Word Flow sure felt cool, but in reality I was typing slower.
This is likely because when using the predictions, you have to type a character or two and thenÂ look through a list of 4 words to see if it correctly predicted your word. If not, you have to type another character, and then check the 4 words again, etc.
The only case where Word Flow was almost as fast as normal typing was in the second to last test. In that test, I first wrote the sentence using normal one-handed typing, and then deleted it and re-wrote the sentence againÂ using prediction, and it was clear that the system knew what I previously just wrote since the suggestions were almost always correct even without typing a letter. So if you only write a few sentences over and over again, Word Flow might be as fast as normal one-handed typing. However, normal typing still came ahead by two seconds.
What do you think of Word Flow? Have you found it to be useful? Have you timed yourself and typed faster thanks to it? Tell us in the comments below.
About the testing technique:
For the first two tests, I filmed them with my camcorder (hence the precision). I rewrote text messages that I previously sent on my old Windows Phone, thus making it entirely realistic as I typed things I would actually send. For the first test, I typed using predictive text first, and then deleted and rewrote the sentence using normal text. For the second test, I reversed that order and wrote out a different message.
For the remaining tests, I timed them with a stopwatch andÂ I wrote out a few comments I found on an article. I started by typing the sentence without predictions first, and then retyped it using predictions, and then for each consecutive test I reversed the order (predictions first, normal next, etc).